On a mechanism of surface oxide formation near the nucleation threshold